Quectel EC25 bug - interaction between generic/3GPP and QMI connection methods

Dan Williams dcbw at redhat.com
Thu Apr 16 15:12:44 UTC 2020


On Thu, 2020-04-16 at 15:20 +0100, Tim Small wrote:
> Hello,
> 
> We've got some EC25, and found the following bug...
> 
> On all tested EC25 firmware versions.
> 
> With ModemManager 1.10.0 and libqmi 1.22.0  and also with
> ModemManager
> 1.12.6 and libqmi 1.24.8.
> 
> Due to the QMI bug recently described on this list, these modems have
> previously been occasionally running with the generic / 3GPP plugin,
> and
> occasionally with the quectel / QMI plugin.
> 
> Things were working OK, up until the time that the modems were
> switched
> to a different SIM card.
> 
> The modems continuously timed out during the "connecting" phase.
> 
> They appear to have retained the settings from the previous SIM card
> (appears to be stored in non volatile memory), and that remembered
> connection seems to stop the new connection (different SIM / APN).

I'm pretty sure that's how all modems work.  The PDP contexts are in
the modem NVRAM and have nothing to do with the SIM card. That's per
the 3GPP standards and have always been that way.

What might be different is that with LTE default bearer autoconnect
each modem can decide what context to use for the autoconnect. Maybe
the EC25 just uses context #1 and that sets up something in the
firmware that makes it fail when trying to activate context 2 with MM.

Dan

> The following workaround gets it to connect:
> 
> > > AT+CGDCONT?
> 
> << +CGDCONT:
> 1,"IPV4V6","EEM2M","0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0",0,0,0,0
> << +CGDCONT: 2,"IP","giffgaff.com","0.0.0.0",0,0,0,0
> <<
> << OK
> 
> > > AT+CGDCONT=1,"IP",""
> 
> << OK
> 
> > > AT+CGDCONT?
> 
> << +CGDCONT: 1,"IP","","0.0.0.0",0,0,0,0
> << +CGDCONT: 2,"IP","giffgaff.com","0.0.0.0",0,0,0,0
> <<
> << OK
> 
> i.e. undefine the first PDP, and it will then connect via QMI.
> 
> Not sure if this is a modem firmware bug, or MM?  Any thoughts?
> 
> Tim.
> 



More information about the ModemManager-devel mailing list